An Efficient Data Structure for General Tree-Like Framework in Mining Sequential Patterns Using MEMISP

Sequential pattern mining is a relatively new data-mining problem with many areas of applications. One of the challenges is to develop a method that is efficient and scalable especially when the sequence database provided gets larger or the minimum support threshold gets smaller. It has been shown that MEMISP algorithm has outperformed all other algorithms in terms of efficiency and scalability. The algorithm, however, has to posess a high performance characteristic. In this paper, we introduce a general tree-like data structure framework for mining sequential patterns. The experimental result shows that this framework considerably improves the performance of MEMISP.